基于查询子主题分类的多样性搜索评价方法

多样化检索结果的评测通常假设一个查询词包含多个权重各不相同的用户子意图,并在此假设的基础上对检索结果进行评测.虽然大多数已经存在的多样化检索评测方法利用了这些特性对检索结果进行评测,但在评测过程中,它们都忽略了查询子意图的类型信息;而不同类型的查询子意图对信息需求具有不同的特点.首先,通过引入衰减函数对这种特点进行描述,进而对用户子意图的分类方法进行抽象;在此基础上,提出了利用查询子意图类型信息进行多样化检索结果评测的框架,该框架定义了利用查询子意图类型信息进行多样化检索评测的方法应该具有的结构;然后,讨论了在用信息类和导航类作为子意图分类方法的前提下,其对应的衰减函数的形式;最后,在TREC与NTCIR测试集上的实验结果表明了所提出方法的有效性.

Subtopic Taxonomy-Aware Framework for Diversity Evaluation

To evaluate search result diversification, which is supposed to meet different needs behind a same query, a number of evaluation frameworks are proposed. Although most of these frameworks take the probability distribution of information needs(usually called subtopics) underlying a query topic into account, they usually do not consider the subtopic taxonomy information. In this paper, the decay function is first introduced to take the subtopic taxonomy information into account. And then based on the decay function, a novel framework called the subtopic taxonomy-aware(STA) framework is proposed to define the structure that the taxonomy-aware diversity evaluation metrics would have. The decay functions used for the informational and navigational subtopics are also discussed. Experiments based on TREC and NTCIR test collections show the effectiveness of the proposed method.